The overall objective of this project is to improve the quality of images obtained by positron emission tomography (PET) for human studies in clinical nuclear medicine. This will be done by developing a fast and accurate computer method for generating images from basic photon-count data acquired by PET scanners having detectors with time-of-flight (TOF) capability. Data from TOF-PET systems contain additional information that permits better spatial localization of coincidence events, compared to conventional (non-TOF) scanners. TOF-PET scanners have been shown to yield significantly better images, especially for large patients; however, the full benefit of TOF has not yet been achieved in the clinical environment due to the relatively slow reconstruction techniques available for TOF-PET data. The proposed work involves the development and testing of an iterative computer method for statistical image reconstruction that makes specific use of the localized nature of TOF-PET data. The method is called DIRECT, short for Direct Image Reconstruction for TOF, and it involves grouping the TOF-PET data based on a novel combination of angular intervals in data space and voxel-like partitions in image space. The hypothesis is that this method will achieve high quantitative accuracy, combined with high computational efficiency, which is critical for obtaining these quantitative images in the clinical environment. High computational efficiency is needed for human studies, since a large amount of data is collected, the image space is sampled on a fine grid, and many iterations are required to accurately recover the activity levels at all locations in the body. Multi-frame studies (e.g., dual time point imaging, dynamic studies) involve multiple runs of the image reconstruction process for which a fast reconstruction technique is essential. In the DIRECT method, the novel grouping of TOF-PET data leads to high efficiency for many of the reconstruction operations; in particular, this grouping enables the operations of forward-projection and back-projection to be done using efficient Fourier-based methods. Achievement of high accuracy combined with high computational efficiency would be a significant step towards realizing the full potential of TOF-PET in the clinical environment, since in current practice performance is compromised for efficiency in order to complete routine whole-body studies in a practical time. Specific aim 1 is designed to formulate, implement, and investigate the DIRECT method for iterative image reconstruction in TOF-PET, focusing on the core components of the method. Specific aim 2 is designed to formulate, implement, and investigate those components of the DIRECT method that involve compensation for the non-ideal characteristics of measured data, including attenuation, scatter, randoms, and detector normalization. Specific aim 3 involves evaluation of the performance of DIRECT in comparison with other TOF-PET reconstruction techniques. PUBLIC HEALTH RELEVANCE: Positron emission tomography is now well established as a valuable imaging tool for the diagnosis of cancer and other diseases and for the planning and monitoring of treatment. The proposed work involves new methods for computer processing of data from a technically advanced generation of PET scanner; the new computer methods are designed to enable these scanners to reach their full potential, leading to improved accuracy of PET images. The proposed work is relevant to public health, since an improvement in the accuracy of PET images would lead to more accurate diagnosis of cancer and other diseases, more accurate planning of treatment, and more accurate monitoring of the response to therapy, leading in turn to better patient outcomes.